In an attempt to learn a bit more about cross over design and speaker design in general, I've started a fictitious design around budget friendly drivers and am only modeling them to learn what does what.
I will state, I do not know what I do not know. I'm currently enrolled in YouTube university trying to figure this stuff out.
Started with some design and budget constraints. 2-way speaker for a 2.0 channel system only. $60 or less for the woofers, $30 or less for tweeters. Floor standing, 40" tall or so. I expect around 2ft^3 with 0.75" thick material. But thats beside the point.
I've been trying to design this around an 8" woofer and within my budget constraints, primarily Dayton Audio, I keep seeing a big spike in the woofer response above the x-over point. Without going to a $200 woofer, what should I be looking into to figure out why this is there? Is there a name for this phenomenon?
Current choice is the Dayton DC215-8 and in Xsim, its showing a 10db spike at 6Khrz when crossed over at 2Khrz. Why is this spike there? Again, this is only modeled, not measured.
Just trying to learn.
I will state, I do not know what I do not know. I'm currently enrolled in YouTube university trying to figure this stuff out.
Started with some design and budget constraints. 2-way speaker for a 2.0 channel system only. $60 or less for the woofers, $30 or less for tweeters. Floor standing, 40" tall or so. I expect around 2ft^3 with 0.75" thick material. But thats beside the point.
I've been trying to design this around an 8" woofer and within my budget constraints, primarily Dayton Audio, I keep seeing a big spike in the woofer response above the x-over point. Without going to a $200 woofer, what should I be looking into to figure out why this is there? Is there a name for this phenomenon?
Current choice is the Dayton DC215-8 and in Xsim, its showing a 10db spike at 6Khrz when crossed over at 2Khrz. Why is this spike there? Again, this is only modeled, not measured.
Just trying to learn.
It's cone breakup and is filtered out by choosing a steeper slope or using a series or parallel notch. The lower you cross, the more suppressed the peak will become in relation to the normal system response.
Read this https://www.diyaudio.com/community/...signing-crossovers-without-measurement.189847
and apply it to your simulation
Read this https://www.diyaudio.com/community/...signing-crossovers-without-measurement.189847
and apply it to your simulation
Posting your models can help people figure out what you have going on also. Even just a screen grab of the circuit and response is better than nothing. If you post the full model files, someone that uses the same software is usually willing to work with them and give you more specific pointers.
D'oh, I completely forgot to post the pic. Forgive me, its a phone pic of a computer screen, don't have the log-ins and forums across the mobile and desk top platforms.
They do this regardless of price so there’s currently nothing to say you can’t get good sound out of this if you approach it correctly.
Isn't It related to the 'ringing' of a filter at Fc? The type employed It's more of a passband/large notch so no dedicated lowpass to combat breakup.
There's a little more to it. The breakup still happens even when you make it flat so you have to cross it away at some lower point. Besides, it's only a peak on axis. Power wise it may not need to be reduced.
Well, apart from the fact that I seek for Dayton DC215-8 and I see nothing but the paper cone which is the DC200, and I see no reasons to make a two way with a 8" aluminum cone, the Dayton DSA215-8, as it appears to be the only '215', but probably I'm mistaken...
Same for filters...I'm forced to see only LC-CL nets, and everything else seems a deviation from the correct way of makin them, but then, again, probably I'm mistaken.
The RC net for enhancing the highs before the tweeter, is it needed? Is it a self-chosen ( by the program) feature or is it voluntarily seeked?
🤔
Same for filters...I'm forced to see only LC-CL nets, and everything else seems a deviation from the correct way of makin them, but then, again, probably I'm mistaken.
The RC net for enhancing the highs before the tweeter, is it needed? Is it a self-chosen ( by the program) feature or is it voluntarily seeked?
🤔
Series resistance and/or inductance could help to bring down the distortion level of cone break-up. So you could try adding say, 10-22 ohms or 1mH, pretend to listen to it, and see if it makes a difference. Reduced distortion could make a difference to the subjective perception of loudness, in spite of the FR plot, so once you hear it, you'll be able to evaluate the importance of further incremental changes to the sound.
Not trying to be facetious, but that's just how it is with some of these things. Like that saying, in theory there's NO difference between theory and practice, but in practice there is.
Not trying to be facetious, but that's just how it is with some of these things. Like that saying, in theory there's NO difference between theory and practice, but in practice there is.
If only to clarify, cone breakup is not distortion, as in it doesn't produce harmonics from nonlinearity because it is a linear effect. @abstract is referring to when lower frequency distortion harmonics fall on the breakup frequencies, and when they are also loud enough to be heard.
@AllenB OK, lazy wording on my part, but the effects all sort-of tie together, don't they?
What we really need is some kind of Matlab / Simulink style model that really illustrates the mechanics, e.g.: a FEA "pearl necklace" arrangement of masses for a model of a flexible cone, with shock absorbers for the surround etc., along with a coil suspended in a slightly imperfect magnetic field, and connect it all up to a signal stream.
I find myself using 1000 words instead of pictures, and probably make mistakes half the time anyway, just because my own mental models are also fallible.
What we really need is some kind of Matlab / Simulink style model that really illustrates the mechanics, e.g.: a FEA "pearl necklace" arrangement of masses for a model of a flexible cone, with shock absorbers for the surround etc., along with a coil suspended in a slightly imperfect magnetic field, and connect it all up to a signal stream.
I find myself using 1000 words instead of pictures, and probably make mistakes half the time anyway, just because my own mental models are also fallible.
I agree that would be nice to have.
In days gone by, breakup has been too difficult to calculate for casual speaker design or even for general purpose simulators. Vituixcad doesn't simulate breakup and it is entirely understandable. In part at least because it doesn't matter.
By doesn't matter, I mean it's common to identify breakup modes and cross them down heavily.. or at the least with a waveguide to cross right about the first mode (not the big peak) due to the beaming that happens in that vicinity due to cone size.. and that beaming is the other reason not to work right into breakup.
Add to that, that you can show breakup through polar measurements so crossover simulators can help you treat it right.
Fullrange users aren't left out of this either, as they have a kind of quasi waveguide effect also with lower cabinet diffraction owing to lobing.. A cross to a tweeter is typically what makes beaming into a problem.
In days gone by, breakup has been too difficult to calculate for casual speaker design or even for general purpose simulators. Vituixcad doesn't simulate breakup and it is entirely understandable. In part at least because it doesn't matter.
By doesn't matter, I mean it's common to identify breakup modes and cross them down heavily.. or at the least with a waveguide to cross right about the first mode (not the big peak) due to the beaming that happens in that vicinity due to cone size.. and that beaming is the other reason not to work right into breakup.
Add to that, that you can show breakup through polar measurements so crossover simulators can help you treat it right.
Fullrange users aren't left out of this either, as they have a kind of quasi waveguide effect also with lower cabinet diffraction owing to lobing.. A cross to a tweeter is typically what makes beaming into a problem.
From the Frequency response, I think it may actually be the DA215-8 rather than DC series that is being modelled, FWIW.
What if you drop the 8"and go 2x DC160-8 6-1/2" in each box MTM or TMM, they on special under $30ea 😎
The FR looks a lot nicer to play with, still will need a pretty reasonable tweeter crossed over low
The FR looks a lot nicer to play with, still will need a pretty reasonable tweeter crossed over low
Hi,
quickly on the process how its usually done: For best results drivers and number of ways and stuff like this come later in design process. Designing crossover is about the last step building speakers, it comes after you have built the system and measured it to be able to make good crossover. Design starts from stuff like defining the application you need the sound for, then you might have a budget like here its the limiting factor, perhaps looks and size targets to have, prioritize these which is the most important to have a guideline.
For best sound you must think about listening position, listening distance, the room acoustics and so on and figure out which kind of acoustic properties of from the speaker would work best. Also some kind of an idea about what sound qualities need to be there, if there needs to be heavy bass, or if you need pinpoint stereo imaging, how loud it needs to be, or what ever feels relevant.
Now you have enough information to actually figure out what kind of a system would fill out the requirements, what compromises there are and what trade-offs you are comfortable with, given you have prioritized the list you want from the system given the application, size, price and looks and of course audio quality requirement. Figure out if it is doable with the budget. If it is not, then figure out what to compromise, how to reduce price. Now you should have a plan, so buy parts and build it, measure, design a crossover (if it needs one😉, evaluate, rinse and repeat if necessary, enjoy. If possible, make cheap fast prototypes, its the fastest and best way to get good results as long as there is time and budget for it.
If you decide from the get go that it needs to be cheap drivers two way without thinking why and what is it that you actually need and how to get it given real constrains, you might be building exactly the wrong thing and be let down in the end. Which is fine as well, as long as you know it before hand it might fail but want to do it nevertheless for learning and having fun. Knowing a goal, and then trying to hit it is the way to do it. Doing something and hoping it hits somewhere is a game of luck. Imagine best sound, best looks, best what ever you fancy and try to figure out how to reach there with the realities you have, and you'll get there eventually.
Most importantly, as it is a hobby, remember to have fun with it, do what ever feels important/meaningful/relevant to you 🙂
Example, if cheapness was only criterion I would slap Tymphany TC9 to a plumbing tube, or just leave it free air, direct to amp, under 10$ per side and takes about 2 minutes to hook a wire to it and bam, done, best sound for the money. I'd be glad with it in mono. Wouldn't be loud, wouldn't look like anything, but would fill in the design criterion nicely. (edit. oops, prices have doubled or trippled since I looked last, would cost 20-30$ depending where you are at).
Last edited:
Primary issue: you don't actually have a crossover on the woofer. You have the basics of a 12 dB/octave electrical crossover on the tweeter (see below though), but only a frequency response contouring circuit on the woofer. Since the natural roll-off of the woofer is far beyond where you want to cross over, this is not a logical approach. As discussed earlier, the breakup region of the driver is also up there and is causing the peak.
And is that coil 620 mH? That ain't right. 🙂
Sticking with the CircuitBlocks in Xsim initially might help you with basic values/configurations.
Last edited:
A quick 2 kHz 2nd order example based on flat impedance (8 ohm), flat frequency response (ideal) drivers. Your likely crossover component values would normally be in the neighborhood of these, as long as you aren't near the natural roll-off of the drivers or some other inherent response anomaly - which is a whole other discussion about acoustic vs. electrical response.
